Ubitium Raises $14M to Build a Universal Processor for Physical AI
German chip startup Ubitium raised $14M led by Ibex Investors for processors that fuse sensor processing, AI inference, and real-time control on a single chip.
One chip for the physical world
Düsseldorf-based Ubitium has raised $14 million in seed financing led by Ibex Investors, bringing total funding to $18 million since the company's founding in 2024. Investors include OpenOcean, KBC Focus Fund, Runa Capital, Inflection, and Quantum Systems co-founder Florian Seibel, a notable signal given Quantum Systems' drone expertise.
The technical pitch
Ubitium develops processors that combine three jobs inside a single architecture: sensor processing, AI inference, and real-time control. For robots, drones, vehicles, and industrial equipment, these three functions usually live on separate hardware, which costs latency, power, and complexity. Fusing them is a bet that physical AI needs purpose-built silicon, not repurposed data center chips.
The company's first silicon, the UB108, contains 256 processing elements and was manufactured using Samsung's 8-nanometre process, showing the startup is already past the paper stage. Ubitium is now developing the UB410 processor with 4,096 processing elements, targeting engineering samples in 2027. The funding will support chip development, engineering recruitment, and commercial applications across robotics, drones, vehicles, and industrial equipment.
Why it matters now
Every humanoid robot, autonomous drone, and factory automation project eventually hits the same wall: the compute that makes decisions has to sit close to the sensors making observations, and it has to react in real time. General-purpose processors do the job today but waste power and latency. Purpose-built physical AI chips are a young category, and Ubitium is positioning early, with real silicon in hand before the market fully forms.
What this means for your business
- Edge AI is becoming a hardware decision. If your business deploys robots, drones, or automated equipment, the processor inside them increasingly determines capability and cost. Track purpose-built AI chips, they'll define what your next hardware refresh can do. - Latency is a feature. Combining sensing, inference, and control on one chip cuts the milliseconds that matter in physical applications. If you're evaluating automation vendors, ask where the compute lives and how fast the control loop is. - 2027 is the window. Engineering samples targeted for 2027 means commercial deployments in 2028 and beyond. Businesses planning robotics or automation roadmaps should factor purpose-built silicon into their timelines. - Small bets on infrastructure pay off. The companies building the chips, the power systems, and the manufacturing for physical AI are where durable value accumulates. Watch the picks-and-shovels layer of the robotics boom.